Application of thermovision camera in gas delivery stations

Security of endangered strategic areas like gas delivery stations is at the moment almost completely dependent on security officers and visual inspection of the scene. Observation of multiple screens displaying images captured from video cameras installed at the station is a monotonous task, therefore there is a great risk of occurring oversights. Due to the human limitations the quality of such observations is far from being efficient and reliable. In order to increase the quality of detection potential threat an algorithm for automatic detection of potential anomalies like gas leaks and pipe damages. Presented algorithm is based on images acquired from thermovision cameras and is capable of processing up to date. In order to perform gas leaks detection, first an area of pipe is detected. Next step is to perform analysis of gradient changes over time. Trespassers detection is realized by detection of silhouettes, analysis of anthropomorphic features and pattern matching. Both algorithms were implemented and tested using real sequences. The acquired results are promising for future implementations.